Actively Exploited Atlassian Zero-Day Bug Allows Full System Takeover

A remote code execution (RCE) vulnerability in all versions of the popular Confluence collaboration platform can be abused in credential harvesting, cyber espionage, and network backdoor attacks.

Digital bug
Source: The lighwriter via Alamy Stock Photo


A critical security vulnerability in Atlassian Confluence is under active attack, opening servers to full system takeover, security researchers warned.

The bug (CVE-2022-26134) is a command-injection issue that allows unauthenticated remote code execution (RCE), affecting all supported versions of Confluence Server and Confluence Data Center. According to a forensic investigation of two zero-day attacks by Volexity, it can be exploited without needing credentials or user interaction, simply by sending a specially crafted Web request to the Confluence system.

No Atlassian Cloud sites have been impacted.

Confluence is a remote working and corporate workspace suite used for project management and collaboration among teams. As such, it houses sensitive data on projects, specific users, and potentially partners and customers; also, it tends to be integrated with other corporate resources, servers, and systems. A successful exploit would allow attackers to vacuum up data from the platform as well as pivot to burrowing deeper into an organization's network as a prelude to, say, a ransomware attack.

"By exploiting this kind of vulnerability, attackers can gain direct access to highly sensitive systems and networks," Volexity researchers noted.

Researchers have advised administrators to remove external access to their Confluence servers immediately until patches have been applied. In the meantime, Atlassian confirmed in its advisory that has rushed a fix, with patches rolling out towards the close of business ET on June 3.

A spokesperson told Dark Reading that the company has "contacted all potentially vulnerable customers directly to notify them of the fix."

Zero-Day Atlassian Confluence Attacks

During its investigation, Volexity followed the path of attackers in two instances, which was the same in both. To start, the culprits exploited the vulnerability to create an interactive webshell (by writing a malicious class file in memory), which gave them persistent backdoor access to the server without having to write anything to disk.

After that, the firm observed that the threat actors dropped the Behinder implant on the server, which is an open source tool for creating flexible memory-only webshells. It also allows integration with Meterpreter and Cobalt Strike, two tools that are most often used for lateral movement. Meterpreter allows users to fetch various Metasploit modules (i.e., working exploits for known bugs), while Cobalt Strike is a pen-testing tool that's often used by the bad guys to probe for and compromise new targets on the network.

Once Behinder was in place, Volexity found that the adversaries went on to install two additional webshells to disk: China Chopper and a custom file upload shell. China Chopper is a tool that's been around for a decade, which allows attackers to retain access to an infected Web server using a client-side application. The client contains all the logic required to control the target, which makes it very easy to use.

Once this basic infection setup was in place, the attackers ran several commands, including those aimed at reconnaissance (checking the operating system, looking for password repositories); stealing information and user tables from the local Confluence database; and altering Web access logs to remove evidence of exploitation, Volexity said.

While the firm detected two zero-day attacks, it's likely that the activity is more widespread. "Volexity has reason to believe this exploit is currently in use by multiple threat actors and that the likely country of origin of these attackers is China," researchers said.

How to Prevent Confluence Compromise

The best option beyond patching to prevent compromise is simply to disable Confluence Server and Confluence Data Center instances, remove all external access, or use IP address safelisting rules to restrict access to only trusted endpoints, researchers noted. Organizations can also add Java deserialization rules that defend against RCE injection vulnerabilities to their Web application firewalls (WAFs).

It's also important to uncover signs of any compromise, given that an infection can persist beyond patching.

"The presence of a webshell provides an attacker with the ability to maintain access to a compromised system even after a vulnerability like this one has been patched," notes Satnam Narang, senior staff research engineer at Tenable. "We observed the same following exploitation of the ProxyShell vulnerability last year, where attackers implanted webshells onto vulnerable Microsoft Exchange Server instances."

However, "these systems can often be difficult to investigate, as they lack the appropriate monitoring or logging capabilities," Volexity pointed out.

Volexity researchers offered the following advice:

  • Ensure Internet-facing Web services have robust monitoring capabilities and log retention policies to assist in the event of an incident

  • Send relevant log files from Internet-facing web servers to a SIEM or Syslog server

  • Monitor child processes of Web application processes for suspicious processes (in this case, the Python shell is a good example of this)

If past is prologue, it's good to be vigilant on this one: Attackers see Confluence as a popular target, as shown by the mass exploitation of another RCE flaw last fall, in volumes that were large enough to trigger a CISA alert.

"While there are currently no exploitation details or proof-of-concept for this vulnerability, we know from history that attackers relish the opportunity to target Atlassian products like Confluence," Narang tells Dark Reading. "We strongly encourage organizations to review their mitigation options until patches are available."

Greg Fitzgerald, co-founder at Sevco Security, also cautions organizations to take proactive steps to generally prevent zero-day attacks.

“Organizations vulnerable to this exploit cannot simply sit back and assume that this will be resolved through their typical patch management process," he tells Dark Reading. "When Atlassian releases a patch, that will be the first step for most organizations. But while patching vulnerabilities works great for the systems that you know about, the vast majority of enterprises simply don’t know the entirety of their attack surface. This is because maintaining an accurate IT asset inventory in a dynamic environment is exceptionally difficult. Threat actors figured that out a long time ago and work around the clock to exploit it. The first step to combating threats like this one is to establish a continuously updated, accurate inventory of all enterprise assets to serve as a foundational control for your security program.”

This post was updated at 4:45 ET to reflect that the bug is no longer unpatched. 

About the Author(s)

Tara Seals, Managing Editor, News, Dark Reading

Tara Seals has 20+ years of experience as a journalist, analyst and editor in the cybersecurity, communications and technology space. Prior to Dark Reading, Tara was Editor in Chief at Threatpost, and prior to that, the North American news lead for Infosecurity Magazine. She also spent 13 years working for Informa (formerly Virgo Publishing), as executive editor and editor-in-chief at publications focused on both the service provider and the enterprise arenas. A Texas native, she holds a B.A. from Columbia University, lives in Western Massachusetts with her family and is on a never-ending quest for good Mexican food in the Northeast.

Keep up with the latest cybersecurity threats, newly discovered vulnerabilities, data breach information, and emerging trends. Delivered daily or weekly right to your email inbox.

You May Also Like

More Insights